Diet analysis by next-generation sequencing indicates the frequent consumption of introduced plants by the critically endangered red-headed wood pigeon ( janthina nitens) in oceanic island Haruko Ando1, Suzuki Setsuko2, Kazuo Horikoshi3, Hajime Suzuki3, Shoko Umehara3, Miho Inoue-Murayama4 & Yuji Isagi1 1Laboratory of Forest Biology, Division of Forest and Biomaterials Science, Graduate School of Agriculture, Kyoto University, Kyoto 606-8502, 2Department of Forest Genetics, Forestry and Forest Products Research Institute, Tsukuba, Ibaraki 305-8687, Japan 3Institute of Boninology, , Ogasawara, Tokyo 100-2101, Japan 4Wildlife Research Center, Kyoto University, Kyoto 606-8203, Japan

Keywords Abstract Conservation, diet analysis, DNA barcoding, next-generation sequencer, oceanic islands, Oceanic island ecosystems are vulnerable to the introduction of alien species, red-headed wood pigeon. and they provide a for many endangered species. Knowing the diet of an endangered is important for appropriate nature restoration efforts on Correspondence oceanic islands because introduced species may be a major component of the Haruko Ando, Graduate School of diets of some endangered species. DNA barcoding techniques together with Agriculture, Kyoto University, Kyoto 606- next-generation sequencing may provide more detailed information on animal 8502, Japan. Tel./Fax: +81 75 753 6129; diets than other traditional methods. We performed a diet analysis using 48 E-mail: [email protected] fecal samples from the critically endangered red-headed wood pigeon that is Funding Information endemic to the Ogasawara Islands based on chloroplast trnL P6 loop sequences. This study used part of DNA barcode The frequency of each detected plant taxa was compared with a microhistologi- database constructed by JSPS KAKENHI Grant cal analysis of the same sample set. The DNA barcoding approach detected a Number 20248017. This study was also much larger number of plants than the microhistological analysis. Plants that supported by the Sasakawa Scientific were difficult to identify by microhistological analysis after being digested in Research Grant from The Japan Science Society and a Grant-in-Aid for JSPS Fellows. the pigeon stomachs were frequently identified only by DNA barcoding. The results of the barcoding analysis indicated the frequent consumption of intro- Received: 4 June 2013; Revised: 25 July duced species, in addition to several native species, by the red-headed wood 2013; Accepted: 12 August 2013 pigeon. The rapid eradication of specific introduced species may reduce the food resources available to this endangered ; thus, balancing eradication Ecology and Evolution 2013; 3(12): 4057– efforts with the restoration of native food plants should be considered. 4069 Although some technical problems still exist, the trnL approach to next-genera- tion sequencing may contribute to a better understanding of oceanic island doi: 10.1002/ece3.773 ecosystems and their conservation.

Introduction (BirdLife International 2008). The diet of an endangered species is an important issue to consider when restoring Oceanic island ecosystems are a high biodiversity conser- native forests and eradicating introduced species; some vation priority because of their high endemism of fauna introduced species have become essential components of and flora owing to long periods of evolution under highly the current ecosystem, which complicates this issue isolated conditions (Whittaker 2007), but they are vulner- (Kawakami 2008). For example, 90% of the diet of the able to human disturbances, such as deforestation and the endangered Ogasawara buzzard Buteo buteo oyoshim is introduction of invasive species (Kachi 2010; Kawakami thought to be composed of introduced (e.g., the 2010). For example, with regard to bird species, 431 black rat Rattus rattus and the green anole Anolis carolin- endangered species inhabit oceanic islands, such as honey- ensis); therefore, the rapid eradication of these introduced creepers in Hawaii and flightless in New Zealand species may negatively affect the buzzard population

ª 2013 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 4057 This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. NGS Diet Analysis of an Endangered Bird H. Ando et al.

(Kato and Suzuki 2005). Thus, a detailed understanding may be an effective noninvasive approach for studying of the food web, including native and introduced species, the diet of the red-headed wood pigeon. Thus far, dietary is required for the conservation of oceanic island ecosys- analyses of this animal have been carried out mainly by tems. direct observation (Kanto Regional Forestry Office 2005, The red-headed wood pigeon Columba janthina nitens 2006; Kanto Regional Environmental Office 2011) and (Fig. 1) is a critically endangered subspecies endemic to microhistological analysis (Shibazaki and Hoshi 2006). the Ogasawara Islands, a chain of oceanic islands located However, the existing information on the diet of this 1000 km south of the main islands of Japan (Fig. 3). The species is fragmented because of the difficulty of continu- present population of this pigeon is thought to include ous observation and species identification of digested approximately 100 individuals, according to observation food items in pigeon stomachs. Via the DNA barcoding records (Horikoshi 2008), and this species is listed as criti- method, fecal sequencing using a next-generation sequen- cally endangered on the Japanese Red List (Environmental cer (NGS) can provide large amounts of sequence data Agency of Japan 2006). The red-headed wood pigeon is without a cloning step. Thus, the time and cost of analy- thought to be a seedeater (Takano et al. 1995) based on sis are reduced, and more detailed animal diet informa- direct observations (feeding on Elaeocarpus photiniifolius, tion can be collected (Valentini et al. 2009a,b; Pompanon Neolitsea boninensis, Melia azedarach and microcarpa, et al. 2012). Although food DNA in fecal samples is often etc. were recorded), similar to other Columba species degraded, the universal short barcode primer trnL g-h (Gibbs et al. 2001). To maintain the foraging habitat of the (Taberlet et al. 2007) can effectively amplify plant DNA, pigeon, a forest must maintain its species diversity and as shown in previous herbivore diet analyses (Valentini supply seeds throughout the seasons (Kawakami 2008). et al. 2009a; Kowalczyk et al. 2011; Raye′ et al. 2011). However, the native forest of the Ogasawara Islands has The higher resolution resulting from the DNA barcod- been destroyed because of human settlements in the 19th ing approach via NGS in comparison with traditional century and World WarII (Kachi 2010; Kawakami 2010). microhistological analysis has been noted in several previ- Furthermore, several introduced plants, such as Bischofia ous studies (Pompanon et al. 2012). However, there have javanica, Ficus microcarpa, Leucaena leucocephala and been few attempts to compare the two methods using the australis, have expanded their populations and same sample set. Soininen et al. (2009) suggested a com- generate a large abundance of (Ecological Society of parison of the high-resolution DNA barcoding approach Japan 2002; Toyoda 2003; Hata et al. 2010; Tanaka et al. with a microhistological analysis for the dietary analysis 2010). The red-headed wood pigeon may frequently of stomach contents. The resolution difference between consume introduced species to fulfill a lack of native food the two methods is unknown for fecal samples, in which resources, and these species will be eradicated during food items are more degraded than stomach contents. nature restoration projects in the near future. The aims of the present study were to estimate the use- Fecal analysis via a DNA barcoding technique depends fulness of the DNA barcoding approach combined with on a taxon identification system using a standardized NGS for the red-headed wood pigeon fecal analysis, to DNA region (Hebert and Gregory 2005). This method compare the resulting resolution to that of the microhis- tological analysis using identical sample sets, and to deter- mine the current dietary composition (regarding whether this diet includes introduced species) of this pigeon.

Materials and Methods Figure 2 provides a general outline for the dietary analysis of the red-headed wood pigeons analyzed in this study. We conducted both NGS DNA barcoding and microhis- tological analysis using the same sample set and com- pared the results of the two methods. After determining the diet of the pigeons, we performed additional analyses using the NGS DNA barcoding data.

Development of the trnL reference database Samples from 222 seed plant species were collected from Figure 1. Red-headed wood pigeon Columba janthina nitens. Chichijima and in the Ogasawara Islands

4058 ª 2013 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. H. Ando et al. NGS Diet Analysis of an Endangered Bird

Feces collection

DNA extraction Filtering

Amplification of trnL P6loop

Roche 454 sequencing

Micro-histological Reference photos P6loop database DNA barcoding analysis and specimens

Figure 2. General outline of the diet analysis Diet of the pigeon for the red-headed wood pigeon.

40N

27N Chichijima (n = 41)

30N Hahajima (n = 7) Figure 3. Sampling locations of the study. The left map shows the location of the Ogasawara Ogasawara Isls. Islands. In the right map, the gray-colored islands are the sampling sites for Chichijima 0 500 km 0 10 km and Hahajima. The numbers in parentheses are 20N the fecal sample sizes. 130E 140E 150E 142E

(Fig. 3). We covered 93% of the woody seed plants (118 final 10 min extension at 72°C. Cycle sequencing was per- species) described in Toyoda (2003). Other woody and formed with a Big Dye Terminator v1.1 Cycle Sequencing herbaceous plants were collected by searching around the Kit (Applied Biosystems, Foster City, CA) according to pigeon habitat. The samples were stored at À30°C before the standard protocol. The cycle sequencing products the DNA extraction, which was performed using a were visualized by the ABI PRISM 3100 Genetic Analyzer DNeasy Plant Mini Kit (Qiagen, Venlo, Netherlands), (Applied Biosystems). To build the reference database, P6 following the manufacturer’s instructions. The universal loop sequences (Taberlet et al. 2007) were extracted from primer pair c–d (Taberlet et al. 1991) was used for PCR the entire trnL (UAA) intron sequence. The Passiflora ed- amplification of whole chloroplast trnL (UAA) introns ulis sequence was retrieved from GenBank and added to (c: 5′-CGAAATCGGT AGACGCTACG-3′;d:5′-GGGGA the database. TAGAGGGACTTGAA C-3′). Each 10 lL of the total reaction mixture volume contained 5 ng of extracted Fecal sampling and DNA extraction DNA, 5 lLof29 Multiplex PCR Master Mix, and 0.2 lmol/L of each primer pair. The PCR conditions were We collected 48 fecal samples from the red-headed wood as follows: denaturation for 15 min at 95°C; 35 cycles of pigeons in Chichijima and Hahajima from September 2009 30 s at 94°C, 1.5 min at 57°C, and 1 min at 72°C; and a to May 2011 (Fig. 3). Sampling on Chichijima was carried

ª 2013 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 4059 NGS Diet Analysis of an Endangered Bird H. Ando et al. out continuously from September 2010 to May 2011. Of Santa Clara, CA) to confirm that the targeted region these samples, 35 were collected after directly observing sequences were correctly amplified and that short frag- pigeon elimination. The remaining 13 were collected from ments were completely excluded during the purification areas around nests and roosts, considering size, shape, and process. Next, the PCR products were mixed such that contents (mainly crushed seeds). The collected feces were approximately the same number of molecules from each stored at À30°C before DNA extraction. DNA was fecal sample was included in each mix (Pegard et al. extracted from 20 mg of fecal dry weight using a DNeasy 2009). Large-scale pyrosequencing was carried out on a Plant Mini Kit (Qiagen). The remainder of each sample 454 GS Junior Sequencer (Roche Diagnostic) following (more than half) was used for microhistological analysis. the manufacturer’s instructions.

PCR amplification and sequencing from DNA barcoding fecal DNA The resulting sequences were separated into each sample To confirm that the 13 samples collected around the nests by MID tags. The sequences in each sample were assem- and roosts belonged to the red-headed wood pigeon, we bled using the CAP Contig Assembly Program in BioEdit amplified a portion of the mitochondrial COI region software (Hall 1999) for a 98% match. The short sequences (290 bp) with the primer pair PSF (5′-AAC sequences (less than 40 bp) and contigs, which included CCGGCACCCTTCTAGGAGACGA-3′) and PSR (5′-AC fewer than four sequences in each sample, were excluded CAGCTAGAGGTGGATAAACAGTT-3′). The primers from DNA barcoding. The DNA barcoding for each con- were designed to avoid amplifying the mitochondrial tig was carried out using a local BLAST in BioEdit. We DNA of other native bird species in the Ogasawara detected a plant within the P6 loop database that exhib- Islands (e.g., the brown-eared bulbul Hypsipetes amaurotis ited the highest score with low e-values (<1.0e–25) for and scaly thrush Zoothera dauma; The Ornithological Soci- each contig. If two or more taxa are assigned the same ety of Japan 2012) besides the red-headed wood pigeon. score for a given contig, the contig was assigned to the Each 10 lL of total reaction mixture volume contained lowest taxonomic level that included both taxa. 5 ng of extracted DNA, 0.05 lL of ExTaq (Takara), 1 lLof 10 9 ExTaq Buffer, 0.8 lL of dNTPs, and 0.12 lmol/L per Microhistological analysis primer pair. The PCR conditions were as follows: denatur- ation for 2 min at 94°C; 40 cycles of 15 s at 94°C, 30 s at We used the 47 fecal samples remaining after DNA 52°C, and 1 min at 72°C; and a final cycle of 5 min at extractions for microhistological analysis. All of one sam- 72°C. We checked for the presence of a PCR product of ple was used for DNA barcoding because of its small suitable length by electrophoresis on a 1.5% agarose gel. amount. We followed the modified method of Shibazaki The universal primer pair g (5′-GGGCAATCCTGAGC and Hoshi (2006). The fecal samples were washed with CAA-3′) and h (5′-CCATTGAGTCTCTGCACCTATC-3′; water and filtered using a 0.5 mm screen. Any fragments Taberlet et al. 2007) was used to amplify the trnLP6 remaining on the screen were examined under the micro- loop. The forward primer was tagged with a multiplex scope at 20 9 magnification. Fragment identification was identifier (MID, Roche Diagnostic) to identify the result- performed using reference photos, and seed specimens ing sequences from each sample. PCR amplification was collected in the Ogasawara Islands. conducted using a Qiagen Multiplex PCR kit (Qiagen). Each 25 lL of total reaction mixture volume contained Data analysis 20 ng of extracted DNA, 15 lLof29 Multiplex PCR Master Mix, and 0.2 lmol/L of each primer pair. The The discrimination rate for the P6 loop database at the PCR conditions were as follows: denaturation for 15 min three taxonomic levels (species, genus, and family) was at 95°C; 45 cycles of 30 s at 94°C, 1.5 min at 57°C, and calculated, as described by Raye′ et al. (2011). The dis- 1 min at 72°C; and a final cycle of 10 min at 72°C. The crimination rate for the species level (Rs) was calculated PCR products were purified using exo/SAP (exonuclease I by dividing the number of unique sequences by the num- and shrimp alkaline phosphatase, Takara, Shiga, Japan and ber of species in the database. The discrimination rate at Promega, Madison, WI) and a High Pure PCR Products the genus level (Rg) was calculated by dividing the num- Purification Kit (Roche Diagnostic, Basel, Switzerland) ber of genera with one or more unique sequences by the and then quantified using a NanoDrop ND-1000 UV–Vis total number of genera. The rate of discrimination for Spectrophotometer (NanoDrop Technologies, Wilming- the species in each family (Rf) was calculated by dividing ton, DE). The length of the PCR products was checked the number of unique sequences per family by the using a High Sensitivity DNA Kit (Agilent Technologies, number of species in the family.

4060 ª 2013 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. H. Ando et al. NGS Diet Analysis of an Endangered Bird

After DNA barcoding, we calculated the frequency of Table 1. Species discrimination rate (Rf) for 38 families that have more than one species in the Ogasawara Islands database. sequence reads (FR) and the frequency of the presence (FP) of each food item. To compare the resolution of Family N species N sequences Rf (%) DNA barcoding and microhistological analysis, the num- bers of detected food plants per sample and frequency of Agavaceae 3 2 67 each food item (number of samples in which the specific Anacardiaceae 2 2 100 Apocynaceae 2 2 100 food item was observed divided by the total number of Aquifoliaceae 3 1 33 samples) were calculated. Each food plant was classified 9 8 89 as native or introduced. Using the data from DNA Caprifoliaceae 2 2 100 barcoding, the following analyses were also performed. To Cyperaceae 12 8 67 estimate the monthly change in diet during the breeding Ebenaceae 2 1 33 season in Chichijima, the relative frequency of data for Euphorbiaceae 5 5 100 major food items during each month was calculated. To Fabaceae 11 11 100 Lauraceae 7 1 14 compare the diets of pigeons on Chichijima and Hahaj- Liliaceae 2 2 100 ima, we performed island-based (samples were pooled for Malvaceae 3 2 50 each island) and sample-based analyses. In the island- Melastomataceae 2 1 50 based analysis, the frequency of reads and the relative Moraceae 9 2 20 frequency of presence data for each plant taxa were calcu- Myrsinaceae 3 2 67 lated. The diet composition differences were tested by Myrtaceae 6 3 50 Pearson’s chi-square tests. In the sample-based analysis, Nyctaginaceae 3 3 100 Oleaceae 3 3 100 nonmetric multidimensional scaling (NMDS) on the Palmae 9 3 33 Chao similarity index (Chao et al. 2004) and an analysis Pandanaceae 2 2 100 of similarities (ANOSIM, Clarke 1993) on a Chao matrix Piperaceae 2 2 100 (Doi and Okamura 2011) were performed by the vegan Pittosporaceae 4 1 25 package in R (Oksanen et al. 2010) using the number of Poaceae 24 17 71 reads and the presence/absence data of each plant taxa. Ranunculaceae 2 2 100 We also compared the diet composition and diversity Rosaceae 4 2 50 Rubiaceae 13 12 92 (number of reads vs. presence/absence of each plant taxa) Rutaceae 9 4 44 measurements in these analyses. Sapindaceae 2 2 100 Sapotaceae 3 1 33 Solanaceae 3 3 100 Results Stachyuraceae 2 1 50 Symplocaceae 2 1 50 Plant discrimination using the trnL P6 loop Theaceae 2 2 100 database Ulmaceae 2 2 100 Urticaceae 2 2 100 The trnL P6 loop was sequenced for 222 plant species Verbenaceae 7 5 75 belonging to 175 genera and 76 families (accession num- Zingiberaceae 3 3 100 bers: AB817341-AB817701). Of the 222 sequences of the P6 loop database, 167 unique sequences were found. The sequence fragment lengths ranged from 66 to 148 bp. The sequencing of 48 fecal samples yielded 35,666 reads, The discrimination rate at the species level, or Rs, was corresponding to an average of 743 Æ 338 (SD) bp and 75%, and the Rg, or genus level, was 89%. All families ranging from 157 to 1747. From the results of the DNA were identified using the P6 loop. Discrimination rates barcoding with the P6 loop database, 44 plant taxa were within families ranged from 14% to 100% (Table 1 and detected from 32,291 reads (approximately 90% of the see Appendix S1 for the correspondence between species total sequences). Most of the reads that were not assigned and groups of species). to a specific plant taxa in the P6 loop database (3225 reads) were short or had a low frequency. Of the 32,391 reads, more than 70% belonged to Morus australis Resolution comparisons between DNA (36.58%) and Gr. Lauraceae1 (34.94%; Table 2), indicat- barcoding and microhistological analysis ing their high consumption by pigeons and/or the high All 13 samples collected around the nests and roosts were PCR amplification efficiency of these plants. The number confirmed as originating from the red-headed wood of detected food plants per sample in the DNA barcoding pigeon by amplification using the primers PSF and PSR. (6.73 Æ 2.70) was significantly greater than that obtained

ª 2013 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 4061 NGS Diet Analysis of an Endangered Bird H. Ando et al.

Table 2. List of the lowest taxonomic levels in the diet of the red-headed wood pigeon and its presence in DNA barcoding and microhistological analysis.

Food items Native/Introduced N reads FS (%) FO (%) DNA barcoding FO (%) microhistology

Plants Morus australis Introduced 11,810 36.58 95.83 0.00 Gr. Lauraceae1 Native 11,280 34.94 70.83 59.57 Gr. Ficus1 Native/Introduced 3902 12.08 93.75 4.00 Fagara boninsimae Native 1328 4.11 33.33 31.91 Gr. Poaceae2 Introduced 922 2.86 66.67 0.00 Ligustrum micranthum Native 514 1.59 6.25 0.00 Leucaena glauca Introduced 335 1.04 12.50 0.00 Sambucus javanica Introduced 297 0.92 20.83 0.00 Carex hattoriana Native 231 0.72 2.08 0.00 Ardisia sieboldii Native 226 0.70 18.75 0.00 Trema orientalis Native 212 0.66 4.17 0.00 Gr. Planchonella1 Native 158 0.49 12.50 12.77 Poaceae Introduced 115 0.36 31.25 0.00 Moraceae Native/Introduced 114 0.35 14.58 0.00 Gr. Palmae2 Native/Introduced 90 0.28 31.25 0.00 Elaeocarpus photiniifolius Native 87 0.27 10.42 2.00 Lantana camara Introduced 77 0.24 18.75 6.38 Buxus liukiuensis Introduced 72 0.22 6.25 0.00 Rhaphiolepis wrightiana Native 67 0.21 8.33 0.00 Solanum nigrum Introduced 63 0.20 2.08 0.00 Juniperus taxifolia Native 62 0.19 4.17 0.00 Calaphyllum inophyllum Native 51 0.16 10.42 0.00 Sonchus oleraceus Introduced 43 0.13 4.17 0.00 Schima mertensiana Native 41 0.13 10.42 0.00 Eurya boninensis Native 23 0.07 6.25 0.00 Celtis boninensis Native 21 0.07 10.42 0.00 Derris elliptica Introduced 20 0.06 6.25 0.00 Hedyotis grayi Native 18 0.06 2.08 0.00 Rutaceae Native 17 0.05 4.17 0.00 Paederia scandens Introduced 13 0.04 4.17 0.00 Livistona chinensis Native 10 0.03 4.17 2.00 Gr. Myrtaceae1 Native/Introduced 10 0.03 2.08 0.00 Scaevola sericea Native 9 0.03 4.17 0.00 Carica papaya Native 9 0.03 2.08 0.00 Pinus luchuensis Introduced 7 0.02 2.08 0.00 Terminalia catappa Native 7 0.02 2.08 0.00 Osmanthus insularis Native 6 0.02 4.17 0.00 Pennisetum purpureum Introduced 5 0.02 2.08 0.00 Drypetes integerrima Native 4 0.01 2.08 0.00 Elaeagnus rotundata Native 4 0.01 2.08 0.00 Melastoma tetramerum Native 4 0.01 2.08 0.00 Paspalum conjugatum Introduced 4 0.01 2.08 0.00 Distylium lepidotum Native 0 0.00 0.00 2.00 Wikstroemia pseudoretusa Native 0 0.00 0.00 2.00 Animals Plumonata Native/Introduced –– 0.15 Arthropoda Native/Introduced –– 0.02

from the microhistological analysis (1.42 Æ 0.62, frequently observed using both methods with similar P < 0.01, t-test). When comparing food items detected by frequencies of presence (Table 2). However, plants such DNA barcoding and microhistological analysis, Lauraceae as Morus australis, Ficus, and Poaceae were frequently (identified as Neolitsea aurata in the microhistological observed using DNA barcoding only. Although they were analysis), Fagara boninsimae, and Planchonella were identified only at low frequencies (observed in one sample

4062 ª 2013 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. H. Ando et al. NGS Diet Analysis of an Endangered Bird each), Distylium lepidotum and Wikstroemia pseudoretusa Discussion were found only by using microhistological analysis. In addition, the shells of snails (Pulmonata) and arthropods, Applicability of DNA barcoding for diet which were not targeted by the DNA barcoding, were analysis of the red-headed wood pigeon observed by microhistological analysis. The DNA barcoding method via trnL P6 loops produced a much higher resolution than microhistological analysis, Monthly change of diet composition and as reported by some previous studies (Soininen et al. difference between the islands, as 2009; Valentini et al. 2009a,b; Raye′ et al. 2011). The estimated by DNA barcoding comprehensiveness of the P6 loop database used in the Figure 4 shows the monthly change in major food plants present study appeared to be sufficient given the high rate (detected in more than 10% of the samples) during the of food plant identification from the pigeon feces. This breeding season in Chichijima. The native species Fagara finding may be explained by the high endemism and low boninsimae and Planchonella were frequently observed species richness of the oceanic island flora, which may only during specific months (December and September, improve the comprehensiveness of the local sequence respectively), whereas native Lauraceae, introduced Morus database and can be a major limitation for food identifi- australis, and Gr. Poaceae2 were consistently observed at cation using the DNA barcoding approach (Valentini high frequencies throughout the breeding season. There et al. 2009a). The results of the present study indicate the were significant differences between the dietary composi- clear advantage of DNA barcoding over microhistological tions of pigeons on Chichijima and those on Hahajima in analysis, the latter of which has a bias favoring easily both of the estimations (frequency of reads and relative identified food items (Soininen et al. 2009). Lauraceae, frequency of presence data; Fig. 5). The results of the Fagara boninsimae, and Planchonella, which were frequently similarity analysis by NMDS and ANOSIM demonstrated observed using both methods, have hard and large seeds. the significant differences between the diet compositions Thus, these plants may be easy to identify using microhisto- on Chichijima and Hahajima based on the presence/ logical analysis because larger fragments may remain in absence data for each plant taxa in each sample feces, even after being crushed in a pigeon’s stomach. In (R = 0.2236, P < 0.05, Fig. 6), and there was no signifi- contrast, Morus australis, Ficus, and Poaceae, which were cant difference according to the analysis based on the frequently observed only by DNA barcoding, have soft and number of reads. small seeds and thus may be difficult to identify using

(A) Moraceae 100% Gr. Palmae2 Gr. Ficus1 80% Lantana camara Sambucus javanica Gr. Poaceae2 60% Morus australis Schima mertensiana 40% Elaeocarpus photiniifolius Gr. Planchonella1 20% Ardisia sieboldii Fagara boninsimae Gr. Lauraceae1 0% (B) 100%

80% Figure 4. Monthly relative frequencies of presence data for major food plants during the 60% Native/Introduced breeding season on Chichijima, based on DNA Introduced barcoding (A). The numbers in parentheses 40% Native show the sample sizes for each month. Each plant sequence group is designated as a native 20% species, as an introduced species, or as containing both native and introduced species 0% (B). Sep. (4) Oct. (2) Nov. (5) Dec. (11) Jan. (10) Feb. (2)

ª 2013 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 4063 NGS Diet Analysis of an Endangered Bird H. Ando et al.

Gr. Lauraceae1 (A) Chichijima I. (Number of reads) (B) Hahajima I. (Number of reads) Morus australis Gr. Ficus1 Fagara boninsimae Gr. Poaceae2 Ligustrum micranthum Carex hattoriana Ardisia sieboldii Gr. Planchonella1 Leucaena glauca Moraceae Trema orientalis Poaceae Elaeocarpus photiniifolius Gr. Palmae2 Rhaphiolepis wrightiana Juniperus taxifolia Sambucus javanica P < 0.001 Schima mertensiana Lantana camara Eurya boninensis Derris elliptica (C) Chichijima I. (Presence) (D) Hahajima I. (Presence) Celtis boninensis Hedyotis grayi Rutaceae Calaphyllum inophyllum Gr. Myrtaceae1 Livistona chinensis Carica papaya Scaevola sericea Pinus luchuensis Osmanthus insularis Paederia scandens Sonchus oleraceus Drypetes integerrima Melastoma tetramerum Paspalum conjugatum Buxus liukiuensis Gr. Rosaceae1 P < 0.001 Elaeagnus rotundata

Figure 5. Comparison of diet composition between Chichijima Island and Hahajima Island based on the frequency of reads (A and B) and the relative frequency of presence data (C and D) for each food item. microhistological analysis; furthermore, most of these Moraceae at 14% and 20%, respectively, Table 1). In the plants are introduced species (except for three native spe- case of Lauraceae, Neolitsea aurata was detected using cies belonging to Gr. Ficus1). These results indicate the microhistological analysis, although this species cannot be underestimation of the pigeons’ use of introduced plants in distinguished from the other Lauraceae species using P6 previous microhistological analyses (Shibazaki and Hoshi loop sequences. The performance of DNA barcoding will 2006). be improved by hierarchical barcoding (Moszczynska However, some limitations of the DNA barcoding et al. 2009) using family-specific barcoding markers. The approach have also been presented. The first problem second problem is that some food items identified by mi- with this technique is its low discrimination rate within crohistological analysis were not observed using DNA the P6 loop database for specific families (Lauraceae and barcoding (Table 2). In the case of the plants Distylium

4064 ª 2013 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. H. Ando et al. NGS Diet Analysis of an Endangered Bird

(A) (B)

Figure 6. The results of the similarity analysis with NMDS and ANOSIM, based on presence/ absence (A) and number of reads (B). White –0.5 0.0 0.5 1.0 stress = 0.13 stress = 0.21 squares designate samples from Chichijima, P < 0.05 –1.0 –0.5 0.0 0.5 1.0 P = 0.71 and black squares designate samples from Hahajima. –0.5 0.0 0.5 1.0 –2 –1 012 lepidotum and Wikstroemia pseudoretusa, there were no including the Japanese wood pigeon Columba janthina sequence mismatches in the primer-binding sites, which janthina (Wild Bird Society of Japan 1998) and the Madeira may have caused the strong amplification efficiency bias laurel pigeon Columba trocaz (Oliveira et al. 2002). The (Deagle et al. 2007). Thus, the inability to detect these fruiting periods of the introduced plants Morus australis plants may have been caused by a failure in the DNA and Poaceae are long and irregular (Toyoda 2003; Ando extraction rather than a failure in the PCR amplification. personal observation), and thus, these plants could be These bias errors can occur when a DNA extraction sam- used throughout the breeding season. Although the P6 ple is obtained from a single scat, particularly for dietary loop database could not identify the introduced plant Ficus analysis of large mammals (Deagle et al. 2005). Although microcarpa or other native Ficus species, the frequently this bias may be small in the fecal analysis of small bird observed sequence for Gr. Ficus1 may belong to the intro- scat, the appropriate extraction strategies (e.g., subsam- duced Ficus microcarpa, considering the observation pling and blending, Deagle et al. 2009; Kowalczyk et al. records of pigeon feeding and the great fruiting abun- 2011) should be considered. The observation of snails and dance of this plant in the Ogasawara Islands (Ecological arthropods in the microhistological analysis indicates that Society of Japan 2002; Shibazaki and Hoshi 2006). Owing the pigeons ate small animals; thus, not only plants were to a lack of sufficient native food resources, the red- targeted in the DNA barcoding method. Diet analysis headed wood pigeon may depend on these introduced using animal-targeted markers (e.g., COI, Meusnier et al. species. 2008) may provide more detailed information regarding In the Ogasawara Islands, the eradication of introduced pigeon ecology. This result also suggests the importance plants has been conducted rapidly over the last decade to of microhistological analysis for understanding the overall conserve the endemic ecosystems (e.g., Bischofia javanica, diet, which is essential for choosing an appropriate barcode Tanaka et al. 2010). Ficus microcarpa has recently been (Pompanon et al. 2012). growing in numbers after the introduction of the pollina- tor Eupristina verticillata (Yokoyama 1996; Ecological Society of Japan 2002) and may also be eradicated in the Monthly changes in diet composition on future as a result of its negative impacts on native ecosys- Chichijima tems (Ecological Society of Japan 2002; Watanabe et al. These results indicate the pigeons’ frequent consumption 2009). To conserve the habitat of the red-headed wood of not only native species but also introduced species. pigeon, the direct and indirect effects of introduced The native Fagara boninsimae and Planchonella species species eradication (Bergstrom et al. 2009; Simberloff seem to be consumed at maximum fruiting times (Toy- et al. 2013) should be considered. The rapid eradication oda 2003). The consistent presence of Lauraceae may be of introduced species may reduce the food resources for explained by the long fruiting period of Neolitsea aurata the pigeon and may have a negative impact on the pigeon (Hayashi 1985) and to the sequence encoded by several population. However, native flora can provide various species that exhibit different fruiting phenologies (e.g., throughout the year, and these plants may recover Neolitsea aurata, Machilus kobu, and Machilus boninensis, after eradicating the introduced species. To facilitate the Toyoda 2003). The results may also indicate the pigeons’ long-term survival of the red-headed wood pigeon, using preference for Lauraceae species. This finding is supported the various native fruits rather than specific dominant by direct observations (Kanto Regional Environmental introduced fruits may be more effective in terms of food Office 2011) and frequent feeding by other (sub)species, quality and quantity. However, the present study could

ª 2013 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 4065 NGS Diet Analysis of an Endangered Bird H. Ando et al. not reveal whether the pigeons selectively use the detected matches in primer-binding sites; Pompanon et al. 2012). food plants or just consumed what is available. An esti- Indeed, one of the shortest plant sequences dominates the mation of food resource availability and quality (e.g., dietary analysis of the alpine chamois Rupicapra rupicapra nutrients) may provide a better understanding of the in Raye′ et al. (2011). Thus, Pompanon et al. (2012) rec- impact of introduced species eradication on the pigeon ommended using the read count data for measuring spa- population. tial or temporal variations in the diet but not for absolute diet quantification. In the case of the present study, com- paring the proportion of each plant read number between Composition differences in pigeon diets the two islands can be reliable (e.g., sequence proportion between the islands of Morus australis). The sequence lengths of the Gr. Laura- Comparing the pigeon food compositions based on the ceae1 (87 bp, only found in Chichijima), Morus australis number of reads and presence data indicated significant (89 bp), and Gr. Ficus1 (89 bp), which exhibited high- differences between Chichijima and Hahajima (Figs. 5, frequency read numbers, were not much shorter than 6A). Although Morus australis and Ficus dominated both those of the other detected food plants (ranging 72- of the islands, the results may indicate different patterns 121 bp, average 89.9 bp), and their frequencies of pres- of food selection by the pigeons on each of the islands. ence were also high. Thus, pigeons may have actually eaten The native Lauraceae species was observed only in these plants in large amounts, as reported by Deagle et al. Chichijima at a high frequency. This finding may be (2010) regarding the accuracy of quantitative diet data explained by its significant fruiting abundance in the for- from the number of reads. However, the pigeon consump- est of Chichijima (Toyoda 2003), indicating its impor- tion of some plants with low PCR efficiency may be tance as a major native food resources on the island. In underestimated in the analysis based on the number of Hahajima, the read percentage of introduced Morus aus- reads. For example, the low sequence proportion of Gr. tralis was much larger than that of Chichijima. These data Poaceae2 in Chichijima (4%) may be explained by its low may indicate the greater usage of this introduced species PCR efficiency, despite its high frequency (75%) (Fig. 5A by the pigeons in Hahajima than those of Chichijima. and C). No significant difference between Chichijima and Leucaena leucocephala, which exhibited the third highest Hahajima in the NMDS analysis was based on the number frequency in terms of both read numbers and presence in of reads, which may reflect the dominance of specific plant Hahajima, is also an expanding introduced species in the sequences (e.g., Morus australis and Gr. Ficus1) through- Ogasawara Islands (Hata et al. 2010). This plant lives on out the samples and the underestimation of some plants the forest edge (Toyoda 2003), which is not the preferred with low PCR efficiency. Another possibility is that the forest habitat of the pigeon. Pigeons on Hahajima may small number of reads for some plants may indicate eat Leucaena leucocephala to make up for a lack of food secondary predation (Sheppard et al. 2005), such as plants resources in the forest area. It is also interesting that the eaten by snails or other invertebrate herbivores eaten by food composition varied among Hahajima samples pigeons, or that the number of reads reflects the biomass (Fig. 6), despite the small sample size and restricted sam- to some extent (Yoccoz et al. 2012). This is an important pling site (five samples were collected in February around issue to be considered in estimating pigeon food prefer- the same roost). The red-headed wood pigeons in ences. Given that appropriate methods to correct various Hahajima are more generalist than those of Chichijima biases in quantitative analysis have never been presented, because of a lack of food resources. Estimations of food it is advisable to use not only the number of reads but also resource availability and temporal diet variation on each the frequency of presence to estimate dietary composition island may reveal differences in pigeon feeding strategies and diversity, as in Soininen et al. (2009), Raye′ et al. between these locations. (2011), and Shehzad et al. (2012). We also compared data from the number of reads and the relative frequency of presence for each plant taxa. The Conclusions reliability of quantitative diet analysis using the sequence count data has been discussed in several previous studies In this study, a diet analysis using DNA barcoding (e.g., Deagle et al. 2009, 2010; Soininen et al. 2009; Valen- provided a high-resolution identification of food plants tini et al. 2009a; Raye′ et al. 2011; Pompanon et al. 2012; and clearly overcame the bias of traditional microhistolog- Shehzad et al. 2012). The PCR efficiency and resulting ical analysis. The results of the DNA barcoding indicated number of sequences from each food item can be affected frequent consumption of introduced species, rather than by several biological (e.g., number of DNA copies in a unit only native species, by the pigeons. The rapid eradication mass of tissues, digestion process) and technical factors of some introduced species without restoration of the (e.g., different sequence lengths among species, mis- native seed plants may reduce available food resources for

4066 ª 2013 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. H. Ando et al. NGS Diet Analysis of an Endangered Bird this pigeon. Thus, a strategy that balances the eradication Chao, A., R. L. Chazdon, R. K. Colwell, and T. J. Shen. 2004. of introduced plants and the restoration of native food A new statistical approach for assessing similarity of species resources is important. Differences between the composi- composition with incidence and abundance data. Ecol. Let. tion of pigeon diets on Chichijima and those on Hahajima 8:148–159. should also be considered during the restoration of each Clarke, K. R. 1993. Non-parametric multivariate analyses island. Although some existing technical problems must of changes in community structure. Aust. J. Ecol. still be solved (e.g., the discrimination rate of the P6 loop 18:117–143. database, sampling strategy), the NGS DNA barcoding Deagle, B. E., D. J. Tollit, S. N. Jarman, M. A. Hindell, A. W. approach will provide a better understanding of the food Trites, and N. J. Gales. 2005. Molecular scatology as a tool web, including the interactions between native and intro- to study diet: analysis of prey DNA inscats from captive – duced species and appropriate nature restoration planning Steller sea lions. Mol. Ecol. 14:1831 1842. for oceanic island ecosystems. Deagle, B. E., N. J. Gales, K. Evans, S. N. Jarman, S. Robinson, R. Trebilco, et al. 2007. Studying seabird diet through genetic analysis of faeces: a case study on macaroni Acknowledgments penguins (Eudyptes chrysolophus). PLoS ONE 2:e831. We thank S. Kaneko, H. Tojyu, S. Yamamoto, and A. Deagle, B. E., R. Kirkwood, and S. N. Jarman. 2009. Analysis Hayano for their support in undertaking various labora- of Australian fur seal diet by pyrosequencing prey DNA in – tory techniques. Regarding the fieldwork, we thank Y. faeces. Mol. Ecol. 18:2022 2038. Hoshi, H. Kato, T. Yasui, C. Takahashi, Y. Watanabe, M. Deagle, B. E., A. Chiaradia, J. McInnes, and S. N. Jarman. Kumamoto, K. Shimada, R. Shimada, and N. Eimura for 2010. Pyrosequencing faecal DNA to determine diet of little their help with sampling. We also thank M. Yamazaki penguins: is what goes in what comes out? Conserv. Genet. – and A. Tanabe for their valuable advice on the data 11:2039 2048. analysis. The Roche GS Junior Sequencer was operated by Doi, H., and H. Okamura. 2011. Similarity indices, ordination, the Cooperation Research Program of the Wildlife and community analysis tests using the software R. Japan J. Ecol. 61:3–20 (in Japanese with English summary). Research Center at Kyoto University. This study used a Ecological Society of Japan. 2002. Handbook of alien species portion of the DNA barcode database constructed by JSPS in Japan. Tanin Shokan, Tokyo, Japan (in Japanese). KAKENHI Grant Number 20248017. This study was also Environmental Agency of Japan. 2006. The Japanese bird red supported by the Sasakawa Scientific Research Grant from list. Environment Agency of Japan, Tokyo, Japan (in The Japan Science Society and a Grant-in-Aid for JSPS Japanese). Fellows. Gibbs, D., E. Barnes, and J. Cox. 2001. Pigeons and doves: a guide to the pigeons and doves of the world. Yale University Data Accessibility Press, New Haven and London, U. K. Hall, T. A. 1999. BioEdit: a user-friendly biological sequence DNA sequences of chloroplast trnL P6 loop database can alignment editor and analysis program for Windows 95/98/ be found in the DDBJ accessions AB817341-AB817701. NT. Nucleic Acids Symp. Ser. 41:95–98. A FASTA file of the fecal sequence will be archived in Hata, K., J. Suzuki, and N. Kachi, 2010. Effects of an alien the Dryad repository. shrub species, Leucaena leucocephala, on establishment of native mid-successional tree species after disturbance in the Conflict of Interest national park in the Chichijima island, a subtropical oceanic island. Pp. 103–110 in K. Kawakami, I. Okochi, eds. None declared. Restoring the oceanic Island ecosystem. impact and management of invasive alien species in the . References Springer, Tokyo, Japan. Bergstrom, D. M., A. Lucieer, K. Kiefer, J. Wasley, L. Belbin, Hayashi, Y. 1985. Woody plants of Japan (Nihon no T. K. Pedersen, et al. 2009. Indirect effects of invasive Jyumoku). Yama-to-keikoku-sha, Tokyo, Japan (in species removal devastate World Heritage Island. J. Appl. Japanese). Ecol. 46:73–81. Hebert, P. D. N., and T. R. Gregory. 2005. The promise of BirdLife International. 2008. Many threatened birds are DNA barcoding for . Syst. Biol. 54:852–859. restricted to small islands. Presented as part of the BirdLife Horikoshi, K. 2008. Final report red-headed wood pigeon State of the world’s birds website. Available at: http://www. Population and Habitat Viability Assessment (PHVA) birdlife.org/datazone/sowb/casestudy/173 (accessed 2 April workshop. Red-headed Wood Pigeon PHVA Workshop, 2013). Ogasawara, Japan (in Japanese).

ª 2013 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 4067 NGS Diet Analysis of an Endangered Bird H. Ando et al.

Kachi, N. 2010. Impact of invasive alien species on native Pegard, A., C. Miquel, A. Valentini, E. Coissac, F. D. R. ecosystems on the Bonin Islands. Pp. 11–14 in Bouvier, D. Francois, et al. 2009. Universal DNA based K. Kawakami, I. Okochi, eds. Restoring the oceanic island methods for assessing the diet of grazing livestock and ecosystem. impact and management of invasive alien species wildlife from feces. J. Agric. Food Chem. 57:5700–5706. in the Bonin Islands. Springer, Tokyo, Japan. Pompanon, F., B. E. Deagle, W. O. C. Symondson, D. S. Kanto Regional Environmental Office. 2011. Report about Brown, S. N. Jarman, and P. Taberlet. 2012. Who is eating conservation of the red-headed wood pigeon in breeding what: diet assessment using next generation sequencing. season (Akagashirakarasubato Hogo Zousyoku Jigyou ni Mol. Ecol. 21:1931–1950. Kansuru Tyousa Tou Gyoumu (Hansyokuki) Houkokusyo). Raye′, G., C. Miquel, E. Coissac, C. Redjadj, A. Loison, and Kanto Regional Environmental Office, Tokyo, Japan (in P. Taberlet. 2011. New insights on diet variability revealed Japanese). by DNA barcoding and high-throughput pyrosequencing: Kanto Regional Forestry Office. 2005. Akagashirakarasubato chamois diet in autumn as a case study. Ecol. Res. 26:265– Kishyo Yasei Doshokubutusyu-shu Hogokanri Taisaku 276. Chosa Hokokmusyo (Report about Conservation and Shehzad, W., T. Riaz, M. A. Nawaz, C. Miquel, C. Poiliot, Management of Endangered Species of Natural Fauna and E. Coissac, et al. 2012. Carnivore diet analysis based on Flora: Red-headed Wood Pigeon). Kanto Regional Forestry next-generation sequencing: application to the leopard cat Office, Tokyo, Japan (in Japanese). (Prionailurus bengalensis) in Pakistan. Mol. Ecol. 21:1951– Kanto Regional Forestry Office. 2006. Akagashirakarasubato 1965. Kishyo Yasei Doshokubutusyu-shu Hogokanri Taisaku Sheppard, S. K., J. R. Bell, K. D. Sunderland, J. Fenlon, D. J. Chosa Hokokmusyo (Report about Conservation and Skirvin, and W. O. C. Symondson. 2005. Detection of Management of Endangered Species of Natural Fauna and secondary predation by PCR analyses of the gut contents of Flora: Red-headed Wood Pigeon). Kanto Regional Forestry invertebrate generalist predators. Mol. Ecol. 14:4461–4468. Office, Tokyo, Japan (in Japanese). Shibazaki, F., and Y. Hoshi. 2006. Japanese Wood Pigeons Kato, Y., and T. Suzuki. 2005. Introduced animals in the diet Columba janthina nitens as a seed disperser in the of the Ogasawara buzzard, an endemic insular raptor in the Ogasawara Islands, southern Japan. Strix 24:171–176 Pacific Ocean. J. Raptor Res. 39:173–179. (in Japanese with English summary). Kawakami, K. 2008. Threats to indigenous biota from Simberloff, D., J. L. Martin, P. Genovesi, V. Maris, D. A. introduced species on the Bonin Islands southern Japan. Wardle, F. Courchamp, et al. 2013. Impacts of biological J. Disaster Res. 3:174–186. invasions: what’s what and the way forward. Trends Ecol. Kawakami, K. 2010. What’s the Bonin Islands? Pp. 3–7in Evol. 28:58–66. K. Kawakami, I. Okochi ed, Restoring the oceanic island Soininen, E. M., A. Valentini, E. Coissac, C. Miquel, L. Gielly, ecosystem. impact and management of invasive alien species C. Brochmann, et al. 2009. Analysing diet of small in the Bonin Islands. Springer, Tokyo, Japan. herbivores: the efficiency of DNA barcoding coupled with Kowalczyk, R., P. Taberlet, E. Coissac, A. Valentini, C. high-throughput pyrosequencing for deciphering the Miquel, T. Kaminski, et al. 2011. Influence of composition of complex plant mixtures. Front. Zool. 6:16. management practices on large herbivore diet—Case of Taberlet, P., L. Gielly, G. Pautou, and J. Bouvet. 1991. European bison in Białowieza Primeval Forest (Poland). Universal primers for amplification of three non-coding For. Ecol. Manage. 261:821–828. regions of chloroplast DNA. Plant Mol. Biol. 17:1105– Meusnier, I., G. A. C. Singer, J. F. Landry, D. A. Hickey, P. D. N. 1109. Hebert, and M. Hajibabaei. 2008. A universal DNA Taberlet, P., E. Coissac, F. Pompanon, L. Gielly, C. Miquel, mini-barcode for biodiversity analysis. BMC Genomics 9:214. A. Valentini, et al. 2007. Power and limitations of the Moszczynska, A., S. Locke, J. D. McLaughlin, D. Marcogliese, chloroplast trnL (UAA) intron for plant DNA barcoding. and T. Crease. 2009. Development of primers for the Nucleic Acids Res. 35:e14. mitochondrial cytochrome c oxidase I gene in digenetic Takano, H., K. Otsu, S.-I. Seki, M. Sato, and T. Toyoda. 1995. trematodes (Platyhelminthes) illustrates the challenge of Ecology and conservation of the Japanese wood pigeon, barcoding parasitic helminths. Mol. Ecol. Res. 9:75–82. Columba janthina nitens, in Ogasawara. Trans. 47th Meet. Oksanen, J., F. G. Blanchet, R. Kindt, P. Legendre, R. B. Kanto Branch Jpn. For. Soc. Pp. 67–70 (in Japanese). O’Hara, G. L. Simpson, et al. 2010. Vegan: Community Tanaka, N., K. Fukuhara, K. Otsu, E. Noguchi, and F. Koike. Ecology Package. R package version 1.17-4. Available from 2010. Eradication of the Invasive Tree Species Bischofia http://cran.r-project.org/, http://vegan.r-forge.r-project.org/. javanica and Restoration of Native Forests on the Ogasawara Oliveira, P., P. Marrero, and M. Nogales. 2002. Diet of the Islands. Pp. 161–171 in K. Kawakami, I. Okochi, eds., endemic Madeira Laurel Pigeon and fruit resource Restoring the oceanic island ecosystem. impact and availability: a study using microhistological analyses. Condor management of invasive alien species in the Bonin Islands. 104:811–822. Springer, Tokyo, Japan.

4068 ª 2013 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. H. Ando et al. NGS Diet Analysis of an Endangered Bird

The Ornithological Society of Japan. 2012. Check-list of Seitai to Hogo ni kansuru Chosa). Tokyo Metropolitan Japanese birds, 7th edn The Ornithological Society of Japan, Government, Tokyo, Japan (in Japanese). Sanda, Japan. Yoccoz, N. G., K. A. Brathen, L. Gielly, J. Haile, M. E. Edwards, Toyoda, T. 2003. Flora of the Bonin Islands, 2nd edn. T. Goslar, et al. 2012. DNA from soil mirrors plant taxonomic Aboc-sha, Kamakura, Japan (in Japanese). and growth form diversity. Mol. Ecol. 21:3647–3655. Valentini, A., C. Miquel, M. A. Nawaz, E. Bellemain, and E. Yokoyama, J. 1996. The occurrence of Eupristina (Parapristina) Coissac. 2009a. New perspectives in diet analysis based on verticillata WATERSTON (Hymenoptera, Agaonidae) in the DNA barcoding and parallel pyrosequencing: the trnL Bonin Islands. Jap. J. Entomol. 64:91–92. approach. Mol. Ecol. Resour. 9:51–60. Valentini, A., F. Pompanon, and P. Taberlet. 2009b. DNA Supporting Information barcoding for ecologists. Trends Ecol. Evol. 24:110–117. Watanabe, K., H. Kato, and T. Sugawara. 2009. Conservation Additional Supporting Information may be found in the of the endemic species on the Ogasawara Islands Psychotria online version of this article: homalosperm. Ogasawara Kenkyu Nenpo 32:11–26. Appendix S1. Whittaker, J. R. 2007. Island biogeography, 2nd edn. Oxford Species groups representing the lowest tax- Univ. Press, Oxford, U. K. onomic levels in the P6-loop database and their associ- Wild Bird Society of Japan. 1998. Research on the ecology and ated families. conservation of the Japanese wood pigeon (Karasubato no

ª 2013 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 4069